Determination of Primary and Secondary Lahar Flow Paths of the Fuego Volcano (Guatemala) Using Morphometric Parameters

remote sensing

Article Determination of Primary and Secondary Lahar Flow Paths of the Fuego Volcano (Guatemala) Using Morphometric Parameters

Marcelo Cando-Jácome and Antonio Martínez-Graña * Geology Department, External Geodynamics Area, Faculty of Sciences, University of Salamanca, Plaza Merced s/n, 37008 Salamanca, Spain; [email protected] * Correspondence: [email protected]; Tel.: +34-923-294496; Fax: +34-923-923294514

Received: 2 February 2019; Accepted: 25 March 2019; Published: 26 March 2019

Abstract: On 3 June 2018, a strong eruption of the Fuego volcano in Guatemala produced a dense cloud of 10-km-high volcanic ash and destructive pyroclastic flows that caused nearly 200 deaths and huge economic losses in the region. Subsequently, due to heavy rains, destructive secondary lahars were produced, which were not plotted on the hazard maps using the LAHAR Z software. In this work we propose to complement the mapping of this type of lahars using remote-sensing (Differential Interferometry, DINSAR) in Sentinel images 1A and 2A, to locate areas of deformation of the relief on the flanks of the volcano, areas that are possibly origin of these lahars. To determine the trajectory of the lahars, parameters and morphological indices were analyzed with the software System for Automated Geoscientific Analysis (SAGA). The parameters and morphological indices used were the accumulation of flow (FCC), the topographic wetness index (TWI), the length-magnitude factor of the slope (LS). Finally, a slope stability analysis was performed using the Shallow Landslide Susceptibility software (SHALSTAB) based on the Mohr–Coulomb theory and its parameters: internal soil saturation degree and effective precipitation, parameters required to destabilize a hillside. In this case, the application of this complementary methodology provided a more accurate response of the areas destroyed by primary and secondary lahars in the vicinity of the volcano.

Keywords: volcano deformation; lahars hazard; magma accumulation; pyroclastic ﬂows; ash plumes

1. Introduction Primary lahars are flows which are formed as a direct result of a volcanic eruption. They tend to be bulky (107–109 m3) and record high speeds (>20 m/s). Their maximum flows are commonly between 103–105 m3/s. These features provide the ability to flow long distances, even hundreds of kilometers downstream. They occur primarily when during an eruptive event incandescent material causes the fast melting of large volumes of ice and snow of the glaciers that cover some volcanic edifices and flows by descent gullies. Secondary lahars mainly include lahars caused by the rains. The unbound, by previous eruptions, pyroclastic material can be easily removed by the rains. In general, these are at lower speed, volume and they travel shorter distances as compared to primary lahars, however, they are most frequent during periods of rain [1]. Guatemala is the only country in the Central American region that has trained local observers to generate reports on volcanic activity, who then transmit the information via radio and/or telephone three times a day. The information is transmitted to the National Institute of Seismology, Volcanology, Meteorology and Hydrology (INSIVUMEH) plant when the following characteristics are present: changes in the release of energy, increase in the number of explosions, increase in the expulsion of ash, increase in seismic activity, rumblings, shock waves, and the manifestation of block avalanches that descend through the ravines in the volcanic perimeter.

Remote Sens. 2019, 11, 727; doi:10.3390/rs11060727 www.mdpi.com/journal/remotesensing Remote Sens. 2019, 10, x FOR PEER REVIEW 2 of 23

43 expulsion of ash, increase in seismic activity, rumblings, shock waves, and the manifestation of block 44 avalanches that descend through the ravines in the volcanic perimeter. 45 On 3 June 2018, a strong eruption of the Fuego volcano in Guatemala produced a dense volcanic 46 ash cloud rising to 10 km high. Following the collapse of the volcanic column, pyroclastic flows and 47 descending lahars caused significant damage and a high number of fatalities around the mountain Remote Sens. 2019, 11, 727 2 of 20 48 [2]. After the eruption, INSIVUMEH, with support from the Volcano Disaster Assistance Program 49 (VDAP) of the United States Geological Survey (USGS), the University of Edinburgh, and Michigan 50 TechnologicalOn 3 June University, 2018, a strong elaborated eruption 2 scenarios of the Fuegoof lahars volcano for medium in Guatemala and heavy produced rains based a denseon the 51 volcanicnumerical ash models cloud risingobtained to 10from km the high. program Following LAHAR the collapseZ (Figure of 1). the These volcanic scenarios column, can pyroclastic be seen on 52 ﬂowsthe MapAction and descending page [3]. lahars caused signiﬁcant damage and a high number of fatalities around the 53 mountainOn 19 [2 November]. After the and eruption, for the INSIVUMEH, fifth time during with 2018, support the fromFuego the volcano Volcano erupted, Disaster with Assistance a 1,200- 54 Programmeter long (VDAP) lava flow of the that United descended States Geologicalinto the Ceniza Survey canyon. (USGS), This the volcanic University flow of with Edinburgh, fragments and of 55 Michiganrock producws Technological by erosion University, on the slopes elaborated of the 2volcano scenarios moved of lahars downhill, for medium incorporating and heavy enough rains basedwater, 56 onso thethat numerical they formed models mud obtained flows and from volcanic the program debris or LAHAR lahars Zthat (Figure descended1). These the scenarios slopes of canthe bevolcano, seen 57 onaffecting the MapAction sectors such page [as3]. Las Lajas, El Jute, besides the Ceniza and Seca ravines, according to 58 INSIVUMEH in its last special volcanological bulletins (Nos. 157-2018 and 158-2018) [4,5].

61 FigureFigure 1. 1.Web Web Map Map prepared prepared by by MapAction MapAction as as a collaboratora collaborator on on the the HazMap HazMap project project and and in supportin support of 62 Nationalof National Institute Institute of Seismology, of Seismology, Volcanology, Volcanolo Meteorologygy, Meteorology and Hydrology and Hydrology (INSIVUMEH). (INSIVUMEH). Data from Data 63 preliminaryfrom preliminary maps ofmaps the of hazard the hazard of pyroclastic of pyroclastic flows flows and lahars and lahars for scenario for scenario A (moderate A (moderate rainfall) rainfall) and 64 scenarioand scenario B (very B heavy(very rainfall).heavy rainfall). Prepared Prepared in June 2018in June by INSIVUMEH, 2018 by INSIVUMEH, Volcano Disaster Volcano Assistance Disaster Program (VDAP), United States Geological Survey (USGS), the University of Edinburgh, and Michigan 65 Assistance Program (VDAP), United States Geological Survey (USGS), the University of Edinburgh, Technological University. In C, location of Las Lajas, El Jute, Ceniza and Seca ravines. 66 and Michigan Technological University. In C, location of Las Lajas, El Jute, Ceniza and Seca ravines. On 19 November and for the fifth time during 2018, the Fuego volcano erupted, with a 1200-m 67 The theory of LAHAR Z’s operation will not be studied in this article, since it is not the objective long lava flow that descended into the Ceniza canyon. This volcanic flow with fragments of rock 68 of this study. Its operation is reviewed briefly. In this article, reference is made to the LAHAR Z producws by erosion on the slopes of the volcano moved downhill, incorporating enough water, so that 69 program and its results as a basis for superimposing the results obtained by the Differential they formed mud flows and volcanic debris or lahars that descended the slopes of the volcano, affecting sectors such as Las Lajas, El Jute, besides the Ceniza and Seca ravines, according to INSIVUMEH in its last special volcanological bulletins (Nos. 157-2018 and 158-2018) [4,5]. The theory of LAHAR Z’s operation will not be studied in this article, since it is not the objective of this study. Its operation is reviewed briefly. In this article, reference is made to the LAHAR Z program and its results as a basis for superimposing the results obtained by the Differential Interferometry Remote Sens. 2019, 10, x FOR PEER REVIEW 3 of 23

70 Interferometry Synthetic Aperture Radar Differential (DINSAR) analysis, the morphometric indices Remote Sens. 2019, 11, 727 3 of 20 71 and the unstable areas near the volcano. 72 LAHAR Z was written to delimit areas of potential lahar inundation from one or more user- 73Synthetic specified Aperture lahar Radarvolumes. Differential LAHAR (DINSAR)Z is a code analysis,that is executed the morphometric within a geographic indices and information the unstable system 74areas (GIS), near thewhich volcano. was created by the USGS [4] and is based on a semi-empirical model that delimits flood 75 riskLAHAR zones Z by was lahar written (that is, to areas delimit drawn areas to repres of potentialent the laharprobable inundation floods during from a one lahar or event) more in a 76user-specified digital elevation lahar volumes. model (DEM). LAHAR Z is a code that is executed within a geographic information 77system (GIS),The whichprogram was uses created two by semi-empirical the USGS [4] andequations is based calibrated on a semi-empirical by statistical model analysis that delimitsin the cross 78flood section risk zones of an by area lahar flooded (that is, by areas a lahar drawn (A) toand represent the flooded the probable planimetric floods area during (B) measured a lahar event) in 27 inlahars 79a digitaldeposits elevation of 9 volcanoes model (DEM). in the United States of America, Mexico, Colombia, Canada and Philippines 80 [6]The (Figure program 2). uses two semi-empirical equations calibrated by statistical analysis in the cross 81section of an The area equations flooded are: by a lahar (A) and the flooded planimetric area (B) measured in 27 lahars deposits of 9 volcanoes in the United States of America, Mexico,/ Colombia, Canada and Philippines [6] 82 𝐴=𝛼 𝑉 (1) (Figure2). / 83 The equations are: 𝐵=𝛼𝑉 (2) 2/3 A = α1V (1) 84 Where A is the maximum section area flooded,2/3 B is the total area flooded and V is the volume B = α2V (2) 85 of the lahar, 𝛼 =0.05 and 𝛼 = 200, are constant values. where A is the maximum section area flooded, B is the total area flooded and V is the volume of the lahar, α1 = 0.05 and α2 = 200, are constant values.

86 FigureFigure 2. Diagram 2. Diagram of association of association between between dimensions dimensions of an idealized of an idealized lahar and lahar cross-sectional and cross-sectional (A) and (A) 87 planimetricand planimetric (B) areas calculated (B) areas bycalculated LAHAR by Z forLAHAR a hypothetical Z for a hypothetical volcano. The volcano. ratio of vertical The ratio drop of (H)vertical 88 to horizontaldrop (H) runout to horizontal distance runout (L) describes distance the (L) extent describ ofes proximal the extent volcano of proximal hazard. volcano LAHAR hazard. Z begins LAHAR 89 calculationsZ begins of lahar-inundationcalculations of lahar-inundation hazard zones where hazar a user-speciﬁedd zones where stream a user-specified and the proximal-hazard stream and the 90 zoneproximal-hazard boundary intersect zone [6]. boundary intersect [6].

The LAHAR Z model works under assumptions such as: 91 The LAHAR Z model works under assumptions such as: 921. Floods1. Floods from past from lahars past lahars can provide can provide the information the information basis to basis predict to predict possible possible future floods.future floods. 93 2. Distant hazard zones are confined to the valleys and are directed to the slopes of the volcano. 2. Distant hazard zones are confined to the valleys and are directed to the slopes of the volcano. 94 3. The volume of the proximal lahar controls the extent of the distal flood forming downstream. 3. The volume of the proximal lahar controls the extent of the distal flood forming downstream. 95 4. Very voluminous lahars occur less frequently. 4. Very voluminous lahars occur less frequently. 96 5. No one can predict the size of the next lahar that will descend by a river. 5. No one can predict the size of the next lahar that will descend by a river. Remote Sens. 2019, 10, x FOR PEER REVIEW 4 of 23

97 6. The volume selection for the lahar and the origin site thereof control the flow range. 98 As a complementary mechanism to the results obtained by DINSAR and Lahar-Z, the 99 morphometric model has been used to determine the trajectories of the destructive lahars with greater 100 precision by combining morphometric parameters obtained from the digital elevation model of 12 m 101 of spatial resolution such as the flow accumulation (FCC), the topographic wetness index (TWI), the 102 Remotelength-magnitude Sens. 2019, 11, 727 factor of slope (LS), the degree of internal saturation of the soil (h/z), where4 of z20 is 103 soil depth, h is water level above the failure plane and the effective precipitation (q/T) required to 104 destabilize a slope, where h is the height of the water table and z is the thickness of the colluvium that slides 6. The volume selection for the lahar and the origin site thereof control the flow range. 105 above the failure plane. All these parameters are obtained from the shallow landslide slope stability 106 modelAs (SHALSTAB) a complementary to complement mechanism the to mapping the results of obtainedthis type byof lahars DINSAR using and remote Lahar-Z, sensing the 107 morphometric(differential interferometry, model has been DINSAR) used to determine in Sentinel the images trajectories 1A and of the2A, destructiveto locate areas lahars of withdeformation greater 108 precisionof the relief by combiningon the flanks morphometric of the volcano, parameters areas that obtained are possibly from origin the digital of these elevation lahars. model To determine of 12 m 109 ofthe spatial trajectory resolution of the suchlahars, as parameters the flow accumulation and morpho (FCC),logical the indices topographic were analyzed wetness with index the (TWI), software the 110 length-magnitudeSystem for Automated factor Geoscientific of slope (LS), Analysis the degree (SAGA). of internal The parameters saturation of and the morphological soil (h/z), where indices z is 111 soilused depth, were hthe is wateraccumulation level above of flow the (FCC), failure the plane TWI, and the the length-magnitude effective precipitation factor (q/T) of the required slope (LS). to 112 destabilizeFinally, a slope a slope, stability where analysis h is the was height performed of the waterusing tablethe Shallow and z isLandslide the thickness Susceptibility of the colluvium software 113 that(SHALSTAB) slides above based the failureon the plane.Mohr–Coulomb All these parameters theory and areits parameters: obtained from internal the shallow soil saturation landslide degree slope 114 stability(h/z) and model effective (SHALSTAB) precipitation to complement (q/T), parameters the mapping required of to this destabilize type of lahars a hillside. using remoteIn this case, sensing the 115 (differentialapplication interferometry,of this complementary DINSAR) methodology in Sentinel images provided 1A and a more 2A, to accurate locate areas response of deformation of the areas of 116 thedestroyed relief on by the primary flanks of and the volcano,secondary areas lahars that in are the possibly vicinity origin of the of thesevolcano lahars. [6] To(Figure determine 3). These the 117 trajectoryparameters of theare lahars,described parameters in more anddetail morphological in the Materials indices and were Methods analyzed chapter. with the software System 118 for AutomatedAlthough Geoscientificthe limits of Analysisthe areas (SAGA).delimited The by parametersthe numerical and modeling morphological of lahars indices must used be weretaken 119 thewith accumulation caution and ofare flow considered (FCC), the as references TWI, the length-magnitude and not as absolutes, factor the of comparison the slope (LS). between Finally, the a slopelimits 120 stabilityof hazards analysis by flows was of performed lahars obtained using from the Shallow modeling Landslide with LAHAR Susceptibility Z and the software morphometric (SHALSTAB) model 121 basedis mainly on thedifferentiated Mohr–Coulomb by thei theoryr theoretical and its conceptualization. parameters: internal soil saturation degree (h/z) and 122 effectiveBased precipitation on the damage (q/T), caused parameters by the required pyroclastic to destabilize flows, it a was hillside. found In thisthat case,the volcanic the application hazard 123 ofmaps this and complementary physical vulnerability methodology contained provided errors a more for two accurate main responsereasons: a of lack the areasof field destroyed observation by 124 primarydata to adjust and secondary the limits of lahars the hazard in the map vicinity simula ofted the with volcano LAHAR [6] (Figure Z, and 3the). Thesepoor tracing parameters of possible are 125 describedsecondary in lahars more and detail the in analysis the Materials of their and distal Methods trajectories chapter. originating in proximal sources [5]. 126

127 Figure 3. LAHAR Z Model (red lines) and the poor tracing of possible secondary lahars and analysis 128 Figure 3. LAHAR Z Model (red lines) and the poor tracing of possible secondary lahars and analysis of their distal trajectories originating from proximal sources. The morphometric model combines 129 of their distal trajectories originating from proximal sources. The morphometric model combines parameters such as the ﬂow accumulation (FCC), the topographical humidity index (TWI), and 130 parameters such as the flow accumulation (FCC), the topographical humidity index (TWI), and slopeslope-length factor of the slope to more accurately determine the trajectories of the destructive lahars. 131 length factor of the slope to more accurately determine the trajectories of the destructive lahars. Although the limits of the areas delimited by the numerical modeling of lahars must be taken 132 with2. Materials caution andand areMethods considered as references and not as absolutes, the comparison between the limits of hazards by ﬂows of lahars obtained from modeling with LAHAR Z and the morphometric model is

mainly differentiated by their theoretical conceptualization. Based on the damage caused by the pyroclastic ﬂows, it was found that the volcanic hazard maps and physical vulnerability contained errors for two main reasons: a lack of ﬁeld observation data to adjust the limits of the hazard map simulated with LAHAR Z, and the poor tracing of possible secondary lahars and the analysis of their distal trajectories originating in proximal sources [5]. Remote Sens. 2019, 11, 727 5 of 20

2. Materials and Methods The Sentinel 1 and 2 images obtained from the Alaska Satellite Facility [7], were used to determine the deformations of the terrain using the DINSAR technique [8] and those from Sentinel 2 were used to determine the changes in the morphology of the relief, which in this case was the delimitation of primary and secondary lahars, and compare them with the results obtained from the morphological model. The digital elevation model of 12 m of spatial resolution was obtained from the Copernico Project [9]. The planimetric information such as the results of the LAHAR Z software application was provided by the National Coordinator for Disaster Reduction (CONRED). To complement the results obtained with the DINSAR analysis and the morphological method, in this study, the spatial distribution of potential sources of primary and secondary lahars during eruptions and torrential rains was analyzed using the slope stability model of shallow landslides. SHALSTAB is a plugin of the SAGA GIS. The general flow of processes used in this study can be seen in Figure4. The processes A, B and C are described below: Process (A): DINSAR analysis. The Sentinel 1 and 2 images were obtained from the Alaska Satellite Facility [9]. Sentinel 1 images were used to determine the deformations of the terrain using the differential SAR interferometry (DINSAR) technique [10] and those from Sentinel 2 were used to determine the changes in the morphology of the relief, which in this case was the delimitation of primary and secondary lahars and compare them with the results obtained from the morphological model. Process B: the planimetric information such as the results of the LAHAR Z software application was provided by CONRED. Process C: digital elevation model of 12 m of spatial resolution was obtained from the Copernico Project. The field observations carried out by CONRED technicians and sent in the format of points in shp format, helped to calibrate and verify the results obtained from the presented methodology, as well as to understand which of the three morphological parameters used had greater weight according to the scenario in the one that developed the lahars. For the study of the primary and secondary trajectories of lahars using the DINSAR technique, the morphometric method and the stability analysis, the following parameters were analyzed: The relief deformation analysis by the Sentinel Application Platform (SNAP) [11] followed the workflow indicated in Figure4 from the Sentinel 1B images taken on May 21 and June 14, 2018. This analysis was carried out to determine the possible zones of deformation of the ground where the possible secondary lahars start. Brief synthesis of the processes to obtain the Interferogram:

Base images: Sentinel 1B

S1B_IW_SLC__1SDV_20180521T001317_20180521T001345_011012_0142C5_A797 S1B_IW_SLC__1SDV_20180614T001319_20180614T001347_011362_014DD5_D06F;

Mission: SENTINEL-1B Pass: ASCENDING Polarization: VV

Process diagram in ESA-SNAP [10] (Figure5).

Step 1 Read-Topsar-Split-Orbit ﬁle: Open the products (Figure5A).

S1B_IW_SLC__1SDV_20180521T001317_20180521T001345_011012_0142C5_A797 S1B_IW_SLC__1SDV_20180614T001319_20180614T001347_011362_014DD5_D06F;

View the products. (Figure5B): In the Sentinel-1 IW SLC products, there are 3 subsets IW1, IW2, IW3. Each subset is for an adjacent strip of satellite data collection by TOPS mode. The Fuego volcano can be seen in IW3 (Figure5C). Remote Sens. 2019, 11, 727 6 of 20 Remote Sens. 2019, 10, x FOR PEER REVIEW 6 of 23

146 147 Figure 4. 4. FlowFlow methodological methodological diagram diagram for for the the determin determinationation of primary of primary and andsecondary secondary lahars lahars flow A 148 pathsflow pathsfrom the from following the following processes: processes: (A) Analysis ( ) Analysis of terrain of deformation terrain deformation with differential with differential synthetic synthetic aperture radar interferometry DINSAR, (B) Use of base information of lahars routes 149 aperture radar interferometry DINSAR, (B) Use of base information of lahars routes obtained from obtained from LAHAR Z -National Institute of Seismology, Volcanology, Meteorology and 150 LAHAR Z -National Institute of Seismology, Volcanology, Meteorology and Hydrology - Hydrology—INSIVUMEHIN. (C) Obtaining the digital elevation model of 12m of spatial resolution of 151 INSIVUMEHIN. (C) Obtaining the digital elevation model of 12m of spatial resolution of Copernicus.org, (D) Analysis of Instability of the Relief with Shallow Landslide Susceptibility software 152 Copernicus.org, (D) Analysis of Instability of the Relief with Shallow Landslide Susceptibility (SHALSTAB), (E) Determination of Morphometric Indexes of the Relief with System for Automated 153 software (SHALSTAB), (E) Determination of Morphometric Indexes of the Relief with System for Geoscientific Analyzes (SAGA), (F) Integration of information and determination of possible flow paths 154 Automated Geoscientific Analyzes (SAGA), (F) Integration of information and determination of of primary and secondary lahars (G). 155 possible flow paths of primary and secondary lahars (G).

156 The processes A, B and C are described below:

Remote Sens. 2019,, 1011, x 727 FOR PEER REVIEW 8 7of of 23 20

182

183 Figure 5. (A) The Sentinel Application Platform (SNAP) process used to obtain the deformations of 184 Figure 5. (A) The Sentinel Application Platform (SNAP) process used to obtain the deformations of the relief by DINSAR analysis. Steps (B–F). Sentinel Application Platform (SNAP) passages of each 185 the relief by DINSAR analysis. Steps (B, C, D, E, F). Sentinel Application Platform (SNAP) passages portion of the routine. See text for further explanation. 186 of each portion of the routine. See text for further explanation. Step 2 View a band geocoding-enhanced spectral diversity. To view the data, Intensity_IW3_VV band. 187 Step 1—Read-Topsar-Split-Orbit file: Open the products (Figure 5A). 188 S1B_IW_SLC__1SDV_20180521T001317_20180521T001345_011012_0142C5_A797Coregister the images into a stack: For interferometric processing, two or more images must be 189 coregisteredS1B_IW_SLC__1SDV_20180614T001319_20180614T001347_011362_014DD5_D06F; into a stack. One image is selected as the master and the other images are the slaves. The 190 pixelsView in slave the products. images will (Figure be moved 5B): In to the align Sentinel-1 with the IW master SLC imageproducts, to sub-pixel there are accuracy 3 subsets (Figure IW1, IW2,5D). 191 IW3. Each subset is for an adjacent strip of satellite data collection by TOPS mode. The Fuego volcano 192 canStep be 3 seenForm in IW3 the (Figure Interferogram-Topsar 5C). Deburst-TopoPhase Removal-Filtering-Sarsim Terrain 193 Corection-Write.Step 2—View (Figurea band5E): geocoding-enhanced Interferogram formation spectral from thediversity. InSAR productsTo view menu. the data, 194 Intensity_IW3_VV band. 195 CoregisterThe current the location images of into the a primary stack: For and interfer secondaryometric tabs processing, after the eruptiontwo or more of 3 Juneimages 2018, must were be 196 coregisteredobtained with into a Sentinel a stack. 2One image image of 10is selected m of spatial as the resolution master and where the bandsother images 4, 3 and are 2 werethe slaves. combined The 197 pixelsto obtain in slave the current images position will be ofmoved the lahars to align (Figure with6). the A briefmaster synthesis image to of sub-pixel the processes accuracy to obtain (Figure the 198 5D).real position of lahars is presented below: 199 Step 3—Form the Interferogram-Topsar Deburst-TopoPhase Removal-Filtering-Sarsim Terrain Base Image: Sentinel-2B: 2018-10-22T16: 23: 19.024Z 200 Corection-Write. (Figure 5E): Interferogram formation from the InSAR products menu. Pass: DESCENDING 201 202

Remote Sens. 2019, 10, x FOR PEER REVIEW 9 of 23

203 The current location of the primary and secondary tabs after the eruption of 3 June 2018, were 204 obtained with a Sentinel 2 image of 10 m of spatial resolution where bands 4, 3 and 2 were combined Remote Sens. 2019, 11, 727 8 of 20 205 to obtain the current position of the lahars (Figure 6). A brief synthesis of the processes to obtain the 206 real position of lahars is presented below: 207 Bands:Base Image 4-3-2 : Sentinel-2B : 2018-10-22T16 :23 :19.024Z 208 Pass: DESCENDING Process diagram in ESA-SNAP 209 Bands: 4-3-2 210 StepProcess 1 Read-Open diagram the in products. ESA-SNAP Granule-img-bands (Figure6A). 211 Step 1—Read-Open the products. Granule-img-bands (Figure 6A). 212 ViewView thethe products.products. (Figure(Figure6 6B):B): In In the the Sentine-2B-Granule-Img Sentine-2B-Granule-Img products, products, there there are are 13 13 spectral spectral 213 bandsbands inin thethe visiblevisible andand near-infrarednear-infrared (VNIR)(VNIR) andand short-wavelengthshort-wavelength infraredinfrared (SWIR)(SWIR) spectrum,spectrum, asas 214 showshow inin thethe belowbelow tabletable (Figure(Figure6 C).6C). 215 StepStep 2 Combine 2—Combine bands bands 4-3-2 4-3-2 RGB RGB natural natural colors. colors. Location Location of of primary primary and and secondarysecondary laharslahars 216 (Figure (Figure6D). 6D).

217

218 Figure 6. Process flow in SNAP. Spatial location of primary and secondary lahars from a Sentinel image 219 Figure 6. Process flow in SNAP. Spatial location of primary and secondary lahars from a Sentinel 2. Combination of bands 4, 3 and 2. (A–C) SNAP passages of each portion of the routine in Sentinel 2A 220 image 2. Combination of bands 4, 3 and 2. (A, B, C) SNAP passages of each portion of the routine in images. Location of primary and secondary lahars, (D) Synthesis of final product: limit of the model 221 Sentinel 2A images. Location of primary and secondary lahars, (D) Synthesis of final product: limit of of lahar in red color obtained from LAHAR Z. In lead color the location to that date of the real lahar 222 the model of lahar in red color obtained from LAHAR Z. In lead color the location to that date of the product of the eruption of June 3 and nearby places. 223 real lahar product of the eruption of June 3 and nearby places. Process (D): mass movements analysis. To evaluate the susceptibility for landslides and define 224 Process (D): mass movements analysis. To evaluate the susceptibility for landslides and define potential areas of sediment production through mass removal processes, the SHALSTAB program was 225 potential areas of sediment production through mass removal processes, the SHALSTAB program used, which is a model to map potential shallow landslides. SHALSTAB combines the characteristics 226 was used, which is a model to map potential shallow landslides. SHALSTAB combines the of the stationary subsurface flow (water flow that moves at shallow depths below the surface of the 227 characteristics of the stationary subsurface flow (water flow that moves at shallow depths below the land with a constant velocity), considering the morphology-hydrology relationship [12]. SHALSTAB 228 surface of the land with a constant velocity), considering the morphology-hydrology relationship uses two types of numerical models: the slope stability model and the hydrologic model. 229 [12]. SHALSTAB uses two types of numerical models: the slope stability model and the hydrologic The slope stability model is based on an infinite slope form of the Mohr–Coulomb failure law in 230 model. which the downslope component of the weight of the soil just at failure, τ, is equal to the strength of 231 The slope stability model is based on an infinite slope form of the Mohr–Coulomb failure law in resistance caused by cohesion (soil cohesion and/or root strength), C, and by frictional resistance due 232 which the downslope component of the weight of the soil just at failure, τ, is equal to the strength of to the effective normal stress on the failure plane Equation (3). 233 resistance caused by cohesion (soil cohesion and/or root strength), C, and by frictional resistance due 234 to the effective normal stress on the failureτ = planeC + (σ Equation− µ) tan (3).ϕ (3) 235 𝜏=𝐶( + 𝜎−𝜇) tan 𝜑 (3) where σ is the normal stress, µ is the pore pressure opposing the normal load and tan ϕ is the angle of internal friction of the soil mass at the failure plane. This model assumes, therefore, that the resistances Remote Sens. 2019, 10, x FOR PEER REVIEW 10 of 23

236 237 Where 𝜎 is the normal stress, 𝜇 is the pore pressure opposing the normal load and tan 𝜑 is 238 the angle of internal friction of the soil mass at the failure plane. This model assumes, therefore, that 239 theRemote resistances Sens. 2019 , to11, 727movement along the sides and ends of the landslide are not significant. 9As of 20 240 SHALSTAB’s goal is the mapping of landslide hazards, then establishing zero cohesion maximizes 241 theto degree movement of instability along the possible sides and at the ends study of thesite. landslide By eliminating are not cohesion, signiﬁcant. Equation As SHALSTAB’s (4) can be written goal is 242 as:the mapping of landslide hazards, then establishing zero cohesion maximizes the degree of instability possible at the study site. By eliminating cohesion, Equation (4) can be written as: 243 𝜌𝑔𝑧 cos 𝜃 sin𝜃= 𝜌𝑔𝑧 𝑐𝑜𝑠 𝜃−𝜌𝑔ℎ𝑐𝑜𝑠 𝜃 tan 𝜑 (4) h 2 2 i ρsgz cos θ sin θ = ρsgzcos θ − ρwghcos θ tan ϕ (4)

244 Where z is soil depth, h is water level above the failure plane, cρs and 𝜌𝑤 is the soil and water 245 bulkwhere density,z is soil respectively; depth, h is g water is gravitational level above acceleration the failure plane, and 𝜃 c ρiss andslopeρw angleis the soil(Figure and 7A). water This bulk 246 equationdensity, (5) respectively; can then be g solved is gravitational for h/z which acceleration is the proportion and θ is slope of the angle soil (Figure column7A). that This is saturated Equation at (5) 247 instability:can then be solved for h/z which is the proportion of the soil column that is saturated at instability: 248 h ρs tan θ 249 = 1−= 1 − (5) (5) z ρw tan ϕ 250 ρs h/z could vary from zero (when the slope is as steep as the friction angle) to when the slope 251 h/z could vary from zero (when the slope is as steep as the friction angle) to ρ wwhen the slope is flat (tan θ = 0). An important assumption, however, will be used below which sets a limit on 252 iswhat flat (tan h/z θ= can 0). be.An Itimportant is assumed assumption, that the failurehowever, plane will and be used the shallow below which subsurface sets a flowlimit ison parallel what 253 h/zto can hillslope, be. It is inassumed which that case the h/z failure can onlyplane be and less the than shallow or equal subsurface to 1.0 flow and is any parallel site requiring to hillslope, h/z 254 ingreater which thancase 1h/z is unconditionallycan only be less stable—nothan or equal extreme to 1.0 rain and can any cause site itrequiring to fail. Figure h/z greater7B illustrates than 1 theis 255 unconditionallyrelationship between stable—no h/z and extremetan θ forrain an can angle cause of internal it to fail. friction Figureϕ of7B 45 illustrates◦ and a bulk the density relationship ratio of 256 between1.6. There h/z mayand betan 4 𝜃 stability for an angle states: of any internal slopes friction equal φ to of or 45° greater and a than bulk the density friction ratio angle of 1.6. will There cause 257 maythe be right-hand 4 stability side states: of (3) any go slopes to zero, equal hence to theor greater site is unstablethan the evenfriction if the angle site will is dry cause (h/z the = 0).right- This 258 handscenario side of is “unconditionally(3) go to zero, hence unstable”, the site commonlyis unstable correspondseven if the site to sitesis dry of (h/z bedrock = 0). outcrop.This scenario Because is 259 “unconditionallyh/z cannot exceed unstable”, 1.0 in this commonly model, if correspondstan θ is less thanto sites or of equal bedrock to tan outcrop.ϕ(1-(ρs- ρBecausew) then h/z the cannot slope is 260 exceed“unconditionally 1.0 in this stable”.model, Theif twotan 𝜃 otheris less stability than statesor equal are “stable”to tanφ and(1-(ρ “unstable”,s-ρw) then withthe theslope former is 261 “unconditionallycorresponding to stable”. the condition The two in whichother stability h/z is greater states thanare "stable" or equal and to that "unstable", needed towith cause the instability former 262 corresponding(given by the to right the handcondition side ofin Equationwhich h/z (5)) is greater and the than latter or correspondingequal to that needed to the to case cause in which instability h/z is 263 (givenless than by the that right needed hand to side cause of instability.Equation (5)) and the latter corresponding to the case in which h/z is 264 less than that needed to cause instability. 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 Figure 7. (A) The one-dimensional approximation used in the SHALSTAB slope stability model in which the failure plane, water table, and ground surface are assumed parallel. The slope is θ, the height of the water table is h, and the thickness of the colluvium that slides above the failure plane is z. Typically, the failure plane is at the colluvium-weathered bedrock or saprolite boundary. (B) Definition

of stability ﬁelds. For this example, the angle of internal friction is 45 degrees, and the bulk density ratio is 1.6 [13]. Remote Sens. 2019, 11, 727 10 of 20

Hydrologic model To model the hydrologic controls on h/z. SHALSTAB use a steady state shallow subsurface flow based on the work by [14,15]. Assume that the steady state hydrologic response model mimics what the relative spatial pattern of wetness (h/z) would be during an intense natural storm which is not in steady state. This assumption would break down if precipitation events are sufficiently intense that thin soils on non-convergent sites can quickly reach destabilizing values of h/z before shallow subsurface flow can converge on unchanneled valleys. Figure8 illustrates the geometry and routing of water off the landscape used in the SHALSTAB hydrologic model. If we assume that there is no overland flow, no significant deep drainage, and no significant flow in the bedrock, then q, the effective precipitation (rainfall minus evapotranspiration) times the upslope drainage area, a, must be the amount of runoff that occurs through a grid cell of width b under steady state conditions. Using Darcy’s law, we can write that (6):

qa = ksh cos θ sin θb (6) where sin θb is the head gradient, ks is saturated hydraulic conductivity. At saturation the shallow subsurface flow will equal the transmissivity, T, (the vertical integral of the saturated conductivity) times the head gradient, sinq and the width of the outflow boundary, b and this we can approximate as follows (7): h q a = WI; WI = (7) z T b sen θ where h represents the height of the water table on the sliding surface (m), z the depth of the soil (m), q effective precipitation [mm], T soil transmissivity [m2·day-1], WI (Wetness Index) factor describes the effect of morphology on the subsurface flow (this is calculated in SAGA as the topogrraphic wetness index (TWI)). In order to delimit areas of danger, new studies have applied a methodology with a geographical information system (GIS) base, identifying zones where flows start along the hydrographic network [12]. Morphologically, the triggering of a detritic or lahar flow depends on two factors: the slope of the channel and the critical flow. The speed of an elementary gravity wave is called the critical velocity. When the velocity of the surface water in a channel equals the critical velocity, the elementary gravity waves with upstream motion remain at the point where they were generated and the flow in the channel is called critical. When the superficial velocity of the water in the channel is equal to the critical velocity, the elementary waves of gravity with movement upstream remain at the point where they were generated and the flow in the channel is called critical. Both factors can be obtained from the morphological characteristics of the basin. The first one is evaluated directly from the digital terrain model, while the liquid flow is evaluated indirectly through the drained area [12]. Studies conducted in the Swiss Alps found a relationship for the critical slope (S) that triggered detrital flows and the drained area of the basin (A) (Equation (8)).

S = c A−n (8)

The coefficients c and n have values of 0.32 and 0.20, respectively. The inverse relationship between S and A indicates that a greater slope of the channel will require a smaller liquid flow to trigger a debris flow (a smaller amount of liquid within the flow area). Other authors have proposed classifying lahar flows into four categories: triggering, propagation, deceleration, and deposit [13]. Triggering occurs on slopes greater than the threshold proposed by other authors; 25◦ is known as a slope of minimum probable slope for the initiation of debris flow [13–15], but with a lower angle than the internal friction of the ground and a drained area smaller than 10 km2. The maximum slope Remote Sens. 2019, 10, x FOR PEER REVIEW 12 of 23

333 The coefficients c and n have values of 0.32 and 0.20, respectively. The inverse relationship 334 between S and A indicates that a greater slope of the channel will require a smaller liquid flow to 335 trigger a debris flow (a smaller amount of liquid within the flow area). 336 Other authors have proposed classifying lahar flows into four categories: triggering, 337 propagation,Remote Sens. 2019 deceleration,, 11, 727 and deposit [13]. 11 of 20 338 Triggering occurs on slopes greater than the threshold proposed by other authors; 25 ° is known 339 as a slope of minimum probable slope for the initiation of debris flow [13, 14, 15], but with a lower 340 anglecorresponding than the internal to the internal friction friction of the angleground of and the deposita drained is sucharea that,smaller in areasthan 10 with km2. steeper The maximum slopes, the 2 341 slopeamount corresponding of detritus is to generally the internal modest friction or negligible, angle of the while deposit for drained is such areas that, exceedingin areas with 10 km steeper, the 342 slopes,type of the transport amount becomes of detritus fluvial is generally in place modest of detrital. or negligible, The model while calculates for drained the degree areas of exceeding the internal 10 343 km2,saturation the type of the of transport soil with thebecomes relationships fluvial ofin theplace height of detrital. h/soil The depth model of the calculates groundwater the degree level z; of (h/z) the 344 internal(Equation saturation (9)) is the of necessary the soil with factors the to relationsh destabilizeips the of slopes. the height h/soil depth of the groundwater 345 level z; (h/z) (Equation (9)) is the necessaryh factorsρs to destabilizetanθ the slopes. = ( − ) 346 = (1 − ) 1 (9) (9) z∅ ρw tan∅ 347 This formula allows two limit states to be defined. When h/z is negative, the slope is unstable for 348 This formula allows two limit states to be defined. When h/z is negative, the slope is unstable any degree of saturation and is called “unconditionally unstable”. 349 for any degree of saturation and is called “unconditionally unstable”. On the other hand, when h/z is greater than one, it is called “unconditionally stable” because 350 On the other hand, when h/z is greater than one, it is called “unconditionally stable” because even in cases of saturation, the slope is stable. To define the intermediate situations, it is necessary to 351 even in cases of saturation, the slope is stable. To define the intermediate situations, it is necessary to add the hydrological model. 352 add the hydrological model. The hydrological model considers a subsurface flow in a permanent regime. 353 The hydrological model considers a subsurface flow in a permanent regime. Applying the law of Darcy, on the characteristics of the movement of water through a porous 354 Applying the law of Darcy, on the characteristics of the movement of water through a porous medium, it is possible to arrive at the following expression that links the degree of saturation (h/z) 355 medium, it is possible to arrive at the following expression that links the degree of saturation (h/z) with effective precipitation (q), which is the rainwater that enters the soil and is assimilated by the 356 with effective precipitation (q), which is the rainwater that enters the soil and is assimilated by the plants (Equation (10)). 357 plants (Equation (10)). h q a 358 = 𝑊𝐼;= WI 𝑊𝐼; =WI = (10) (10) z T b sen θ 359 With 360 With 361 q 𝑞= = effective effective precipitation precipitation [mm]; [mm]; 362 T T == soilsoil transmissivitytransmissivity [m[m22·day·day−1];1]; 363 θθ= = slopeslope [[°];◦]; 364 a a == contributioncontribution areaarea [m[m22]; and 365 b b == is thethe lengthlength ofof thethe contourcontour throughthrough whichwhich thethe flowflow transitstransits [m].[m]. 366 367 wherewhere T is theT is transmissivity the transmissivity of the of soil; the soil; a is thea is drainedthe drained area; area; and and b is theb is dimensionthe dimension of the of rasterthe raster cell. 368 cell.The The relation relation q/T q/T represents represents the the magnitude magnitude of theof the precipitation precipitation relative relative to theto the capacity capacity of of the the soil soil to 369 toconduct conduct the the water. water. The The WI WI (wetness (wetness index) index) factor fact describesor describes the the effect effect of groundof ground morphology morphology on theon 370 thesubsurface subsurface flow. flow. 371

372

373 Figure 8. SketchSketch of of hydrological hydrological model: model: ( (AA)) plan plan view view and and ( (BB)) cross cross section section of of a a draining draining area area a a across across 374 a contour of length b. ( (BB)) The The grey grey volume is is the shallow sub- sub-surfacesurface flow flow TMb and the saturation overland flow udb (not calculated by the model). p: precipitation; e: evapotranspiration; r: deep drainage; q: effective rainfall (q = p − e − r); h: vertical depth of the saturated sub-surface flow; z: vertical depth of the failure plane; T: transmissivity; M: sinθ; θ: slope angle. Mettma Ridge study site [13]. Remote Sens. 2019, 11, 727 12 of 20

The analysis of the stability on the slopes of the volcano with SHALSTAB through the hydrological relationship q/T, the magnitude of precipitation, q, in relation to the capacity of the subsoil to transmit water along the T (transmissivity) slopes determined that the greater q was in relation to T, the more likely the soil would be saturated, and the greater the number of sites with unstable slopes. This relationship used the classiﬁcation of pixels with stable positive values to quasi stable and chronically unstable values with negative values according to the stability classes (Table1).

Table 1. Stability classes deﬁned by SHALSTAB and modiﬁed from Montgomery and Dietrich [15].

Classes SHALSTAB Interpretation of Class Chronic instability Unconditionally unstable and unsaturated Log q·T -1 < -3,1 Unconditionally unstable and unsaturated 3,1 < Log q·T -1 < -2,8 Unstable and saturated -2,8< Log q·T-1 < -2,5 Unstable and unsaturated -2,5< Log q·T-1 < -2,2 Stable and unsaturated Log q·T -1 > -2,2 Unconditionally stable and unsaturated Stable Unconditionally stable and saturated

Process E: The morphometric method with SAGA. Once the areas of surface deformation of the relief were found with DINSAR (Sentinel 1 images), the current location of the lahars (Sentinel 2 images), the superficial flow accumulators of the runoff and instability zones (SHALSTAB) were determined, the morphological method was applied to complement the location of trajectories of lahar flows with the analysis of the following morphological index.

2.1. LS-Factor (Slope Length and Steepness Factor) This is a combination of the slope and the length of the slope as when combined, it is a useful attribute to predict erosion potential (Equation (11)). The length factor of the slope, L, calculates the effect of the length of the slope on the erosion, and the slope factor of the slope, S, calculates the effect of the slope on the erosion [16,17]. The higher values of brown color represent a greater susceptibility to erosion (Figure9). LS = Lˆ(1/2)/100*(1.36 + 0.97 + 0.1385*Sˆ2) (11) where L is the length of the slope in meters, and S is the gradient of the slope in %.

2.2. Topographic Wetness Index (TWI) The topographic wetness index models the dynamics of surface and subsurface flows based on the topographic control of runoff, which offers a better perspective in relation to the prediction of sites where the saturation and high concentration of runoff can act as initial flow paths to processes greater than flood (Figure 10). The TWI combines the contribution to runoff from a local area drained and the pending of it, and it is commonly used to quantify topographic control on hydrological processes and is defined as [6–10,18–20] (Equation (12): TWI = Ln (af/tan β) (12) where af is the local area drained for a calculation point, and tan β is the directional slope of the cell of interest (and of the eight neighbors in the case of using an algorithm D8. Ln is natural logarithm. Remote Sens. 2019, 11, 727 13 of 20 Remote Sens. 2019, 10, x FOR PEER REVIEW 14 of 23

402

403 FigureFigure 9.9.LS-factor LS-factor (slope(slope lengthlength andand steepnesssteepness factor)factor) calculatedcalculated ininSAGA SAGAgeographic geographic informationinformation 404 systemsystem (GIS).(GIS). TheThe lowlow susceptibilitysusceptibility toto erosionerosion isis representedrepresented withwith greengreen colorscolors andand thethe highhigh withwith 405 brownbrown color.color. Remote Sens. 2019, 10, x FOR PEER REVIEW 15 of 23 406 2.2. Topographic Wetness Index (TWI) 407 The topographic wetness index models the dynamics of surface and subsurface flows based on 408 the topographic control of runoff, which offers a better perspective in relation to the prediction of 409 sites where the saturation and high concentration of runoff can act as initial flow paths to processes 410 greater than flood (Figure 10).

411 The TWI combines the contribution to runoff from a local area drained and the pending of it, 412 and it is commonly used to quantify topographic control on hydrological processes and is defined as 413 [6–10,18–20] (Equation (12):

414 TWI= Ln (af/tan β) (12) 415 Where af is the local area drained for a calculation point, and tan β is the directional slope of the 416 cell of interest (and of the eight neighbors in the case of using an algorithm D8. Ln is natural 417 logarithm.

418 419 FigureFigure 10. 10.TopographicTopographic wetness wetness index index (TWI) (TWI) calculated inin SAGA SAGA GIS. GIS. Higher Higher values values of TWI of (tendencyTWI 420 (tendencyto blue to color) blue representcolor) represent drainage drainage depressions depr andessions lower and values lower (tendency values (tendency to red color) to representred color) crests 421 representand ridges. crests and ridges.

422 2.3. Flow2.3. Flow Accumulation Accumulation (FCC) (FCC) 423 FlowFlow accumulation accumulation is calculated is calculated as the as theupslop upslopee contributing contributing (catchment) (catchment) area areausing using the theD8 D8 424 modelmodel or multiple or multiple flow flow direction direction approach approach [17–19]. [17– 19The]. Theresult result is an is accumulated an accumulated flow flow raster raster for each for each 425 cell, cell,determined determined by the by accumulation the accumulation of the of weight the weight of all of of all the of cells the cells that thatflow flow into intoeach each cell cellof the of the 426 descendingdescending slope. slope. The Theaccumulated accumulated flow flowis based is based on the on total thetotal number number of cells, of cells, or a orfraction a fraction of them, of them, 427 flowingflowing to each to each cell cellin the in output the output raster. raster. 428 The output cells with a high flow accumulation are areas of concentrated flow and can be used 429 to identify watercourse channels (Figure 11).

430

431

432 Figure 11. Cumulative flow calculated in SAGA GIS. The higher values (tendency to red color) 433 represent greater accumulations of flow and areas of concentrated flow that can be used to identify 434 watercourse channels.

Remote Sens. 2019, 10, x FOR PEER REVIEW 15 of 23

418 419 Figure 10. Topographic wetness index (TWI) calculated in SAGA GIS. Higher values of TWI 420 (tendency to blue color) represent drainage depressions and lower values (tendency to red color) 421 represent crests and ridges.

422 2.3. Flow Accumulation (FCC) 423 Flow accumulation is calculated as the upslope contributing (catchment) area using the D8 424 model or multiple flow direction approach [17–19]. The result is an accumulated flow raster for each 425 cell, determined by the accumulation of the weight of all of the cells that flow into each cell of the 426 descending slope. The accumulated flow is based on the total number of cells, or a fraction of them, 427 Remoteflowing Sens. to2019 each, 11 ,cell 727 in the output raster. 14 of 20 428 The output cells with a high flow accumulation are areas of concentrated flow and can be used 429 to identifyThe output watercourse cells with channels a high ﬂow (Figure accumulation 11). are areas of concentrated ﬂow and can be used to identify watercourse channels (Figure 11). 430

431

432 FigureFigure 11.11. CumulativeCumulative flowflow calculatedcalculated inin SAGASAGA GIS.GIS. TheThe higherhigher valuesvalues (tendency(tendency toto redred color)color) 433 representrepresent greater greater accumulations accumulations ofof flowflow andand areasareas ofof concentrated concentrated flowflow thatthat cancan bebe usedused toto identify identify 434 watercoursewatercourse channels. channels.

3. Results The differential synthetic aperture radar interferometry (DINSAR) analysis, based on the Sentinel 1 images, showed the lack of magmatic deformation before the eruption, but there were surface subsidences on the associated slopes due to the passage and deposits of lahars (Figure 12). This lack of magmatic deformation corroborates the studies undertaken by several researchers in the volcanoes of the area [20–22], which attribute the absence of deformation to differences in the storage of magma in relation to other continental arcs. The absence of magmatic deformation before the lahars can mean that the intrusion of magma is not the trigger of the lahars. Most likely, the rain is the cause. The analysis of stability on the slopes of the volcano with SHALSTAB determined that the zones near the volcano were the most unstable with a qualification of quasi stable to chronic instability through the erosion of the channels where the lahars had flowed and where there had been greater destruction as is the case of El Rodeo, San Miguel, Monte María, and sectors near the Guacalate River (Figure 12B,C). The current locations of the lahars produced by the eruption of 3 June 2018 and later were delimited with the Sentinel 2 images and their destructive power was verified with field data mainly from La Réunion, San José Las Lajas, Santa Clara Las Lajas, San Miguel, and El Rodeo, amongst other towns (Figure 12C). Figure 13A shows the sectors before the eruption. The implementation of lead-colored lahars on that date. In line of red color, the limits of the model of lahars obtained with LAHAR Z. In blue are the ravines and canals where the primary and secondary lahars flowed that flooded the sectors, and in green-yellow-ocher color the areas of ridges and crests. The field control points of the destroyed sites are represented by red crosses. In (B) Sectors close to La Réunion, in (C) sectors near San Miguel and in (D) close to Monte María. Remote Sens. 2019, 10, x FOR PEER REVIEW 16 of 23

435 3. Results 436 The differential synthetic aperture radar interferometry (DINSAR) analysis, based on the 437 Sentinel 1 images, showed the lack of magmatic deformation before the eruption, but there were 438 surface subsidences on the associated slopes due to the passage and deposits of lahars (Figure 12). 439 This lack of magmatic deformation corroborates the studies undertaken by several researchers in the 440 volcanoes of the area [20–22], which attribute the absence of deformation to differences in the storage Remote Sens. 2019, 11, 727 15 of 20 441 of magma in relation to other continental arcs. The absence of magmatic deformation before the lahars 442 can mean that the intrusion of magma is not the trigger of the lahars. Most likely, the rain is the cause.

443

Figure 12. DINSAR analysis and SHALSTAB unstable and erosion areas. (A) Iridescent colors indicate the tendency of deformation in the region in cm. The positive values represent raised areas and the negative subsidence zones with minimal deformation in the volcano on the SE flank. (B) SHALSTAB results: the relief on the flanks of the volcano is very irregular with raised and sunken areas possibly due to the permanent morphological change resulting from the eruptions that have occurred. (C) Detail of the areas near the volcano that are the most unstable, with a rating of almost stable to chronic instability due to erosion of the channels that the lahars have flowed in, have overflowed and caused flooding with the destruction of the civil infrastructure mainly in El Rodeo, San Miguel, Monte María, and sectors near the Guacalate river. Negative red colors and trend towards –10 are unstable areas and stable positive zones with colors towards the blue and values with trend toward 10. The value is dimensionless. Remote Sens. 2019, 11, 727 16 of 20 Remote Sens. 2019, 10, x FOR PEER REVIEW 18 of 23

469

Figure 13. Sequence of images from the upper parts of the volcano of the areas destroyed by the lahars, erosion of channels and floods obtained from the combination of the response of the Sentinel 2A, morphometric index LS, TWI, FlowAcc(acronym of Flow Accumulation), and stabilized in the sectors the day of the eruption of Fuego volcano. (A) The implementation of lead colored lahars on that date. In line of red color, the limits of the model of lahars obtained with LAHAR Z. In blue the ravines and canals where the primary and secondary lahars flowed that flooded the sectors, and in green-yellow-ocher color the areas of ridges and crests. The field control points of the destroyed sites are represented by red crosses. In (B) Sectors close to La Reunion, in (C) sectors near San Miguel and in (D) close to Monte María. Remote Sens. 2019, 10, x FOR PEER REVIEW 19 of 23

470 Figure 13. Sequence of images from the upper parts of the volcano of the areas destroyed by the lahars, 471 erosion of channels and floods obtained from the combination of the response of the Sentinel 2A, 472 morphometric index LS, TWI, FlowAcc(acronym of Flow Accumulation), and stabilized in the sectors 473 the day of the eruption of Fuego volcano. (A) The implementation of lead colored lahars on that date. 474 In line of red color, the limits of the model of lahars obtained with LAHAR Z. In blue the ravines and 475 canals where the primary and secondary lahars flowed that flooded the sectors, and in green-yellow- Remote Sens. 2019, 11, 727 17 of 20 476 ocher color the areas of ridges and crests. The field control points of the destroyed sites are 477 represented by red crosses. In (B) Sectors close to La Reunion, in (C) sectors near San Miguel and in 478 From(D) close the analysisto Monte María. of the morphometric parameters LS, TWI, and FlowAcc, the areas with greater susceptibility to erosion were obtained as well as those with the highest concentration and 479 From the analysis of the morphometric parameters LS, TWI, and FlowAcc, the areas with greater 480 accumulationsusceptibility of superficialto erosion flowwere ofobtained rainwater as thatwell wereas those directly with related the highest to the primaryconcentration and secondaryand 481 laharsaccumulation (Figure 14 of). superficia These areasl flow of of blue rainwater color werethat were observed directly to related have to been the primary directed and by secondary the lahars that 482 destroyedlahars (Figure the populated 14). Theseareas. areas of These blue color three were morphometric observed to have indices been [23 directed–26] are by valid the lahars parameters that to 483 determinedestroyed the the flow populated paths of areas. primary These and three secondary morpho lahars.metric Theindices areas [23–26] with are high valid values parameters of these to indices 484 aredetermine areas of moisture the flow paths concentration of primary in and the secondary slopes, flow lahars. paths The andareas concentration with high values of of sediments these indices that can 485 moveare fromareas of the moisture high parts concentration by intense in the rains slopes, and flow gravity paths generating and concentration secondary of sediments lahars, generallythat can on 486 themove slopes. from the high parts by intense rains and gravity generating secondary lahars, generally on the 487 slopes.

488

489 FigureFigure 14. 14.The The ﬂow flow paths, paths, calculated calculated by using using the the thr threeee morphometric morphometric indices indices TWI, TWI, LS, and LS, FlowAcc and FlowAcc 490 in SAGA,in SAGA, determined determined the the areas areas that that werewere flooded ﬂooded by by primary primary and and secondary secondary lahars. lahars.

491 TheThe ﬁeld field control control was was carried carried outout byby technicians of of CONRED. CONRED.

492 4. Discussion4. Discussion 493 UnlikeUnlike the the LAHARZ LAHARZ model,model, the the presented presented model model uses uses morphometry morphometry as a value as a that value changes that the changes 494 thetrajectory trajectory of of the the primary primary and and secondary secondary lahars, lahars, whic whichh allowed allowed us to us obtain to obtain the real the flow real flowpaths pathsof of 495 thesethese phenomena phenomena and and their their effects. effects. AA beneficialbeneficial coincidence coincidence of ofthe the application application of both of both models models is that is that 496 theythey kept kept a central a central axis axis path path of of thethe mainmain lahars, lahars, due due to to the the fact fact that that they they were were drains drains or main or mainrivers. rivers. 497 However,However, LAHARZ LAHARZ does does not not separate separate secondary secondary lahars, lahars, due due to thisto this action; action; the the morphological morphological method 498 canmethod determine can determine the flow paths the flow of thepaths mentioned of the mentioned phenomenon. phenomenon. 499 The lahar danger map, created for the Fuego volcano, based on the LAHARZ program, The lahar danger map, created for the Fuego volcano, based on the LAHARZ program, published 500 published in 2018 by CONRED, established high hazards zones (marked in red color) that can be in 2018 by CONRED, established high hazards zones (marked in red color) that can be affected by 501 affected by lahar flow paths. Torrential rains have generated secondary lahars that have affected the 502 laharsurrounding flow paths. areas Torrential of the volcano, rains have even generated those ones secondary of low hazard. lahars The that medium have affected hazard thezones surrounding were 503 areasaffected of the by volcano, lahar flow even paths those from ones the ofLa lowReunión hazard. site to The the medium Alsacia site, hazard as can zones be evidenced were affected in Figure by lahar 504 flow13. paths from the La Reunión site to the Alsacia site, as can be evidenced in Figure 13. In the current paper, with the proposed methodology, a study of the lahars flow paths has been carried out, being the main study areas the valleys of these Barranca Seca, Ceniza, Guacalate, Playa Trinidad, Las Cañas, El Jute and Las Lajas rivers. Lahars originated due to the erosion of the cause ways of detritus-ash-water deposited during the volcanic eruption and torrential rains. The axis of the primary lahar paths studied with the proposed methodology are like the ones obtained to delimit the high hazards zones in the mentioned hazard map. On the other side, is has been proved that the modeled secondary lahar deposits with the proposed methodology destroyed civilian structures in La Reunión, San José de Las Lajas, Santa Clara de Las Lajas, San Miguel and other areas had location limits in accordance with the data obtained with LAHARZ and the location obtained with Sentinel 2A images, but with an added element that was not Remote Sens. 2019, 11, 727 18 of 20 obtained with LAHARZ which is the outline of the secondary lahars flow paths, some as contributors to primary lahars and others as independent elements with no connection to the main ones. The range of the secondary lahars and its respective modeled central axis is similar and bigger to the June and November lahars of 2018, observed in the Sentinel images. This is probably due to the resolution of the digital elevation model and the data given by the analyzed morphometric index. Finally, a comparison of the affected areas has been carried out and its reach on transit zones, obtained with the proposed methodology and LAHARZ, observing that the entire area affected by lahars is within the high danger zone. The lahars corresponding to the Las Lajas y Guacalate rivers, take the biggest amount of area of the affected areas, with most of the civilian infrastructure destroyed as a result, turning the mentioned areas to maximum danger zones. The surface terrain deformations, obtained from Sentinel 1B, located in the high mass erosion and removal areas can be part of secondary lahars which were generated from high ground. This is a contribution when the DINSAR analysis is used. In addition to this work, further research studies can improve the location of the lahars flow paths using the deformed areas of the relief obtained with the DINSAR analysis, the stability criterion of slopes and analysis of morphometric parameters in digital models of high elevation spatial resolution obtained from Lidar images and integrating them into specialized programs in lahars and floods analysis such as LAHAR Z, Titan 2D, Ash3D, Flo-2D, IRIC, Iber, among others. The final objective of these investigations is to put in the hands of government authorities and the community, vulnerability maps to this type of threats to reduce them.

5. Conclusions As a contribution of the application of this methodology, based on a flow modeler such as LAHAR Z or another program, to know the Flow Paths of the primary and secondary lahars that have caused the destruction in the vicinity of the volcano, this methodology deepened the study of these trajectories based on the analysis of the morphometric parameters, which allowed us to have a better knowledge of the possible paths before a new volcanic eruption. This methodology is currently being applied in the Fuego volcano by INSIVUMEH technicians with the objective of putting it in the hands of the government authorities, as well as the institutions linked to the subject, which allows these conditions to be ascertained in advance thereby implementing structural measures and non-structural measures to reduce this danger. To complement the results obtained with the DINSAR analysis and the morphological method, in this study, the spatial distribution of the sources and potential routes of primary and secondary lahars during the eruptions and torrential rains that can erode and produce mass movements in areas of critical instability was obtained from the stability model of shallow landslide slopes with SHALSTAB. The integration of the DINSAR methodologies to determine the surface deformations of the ground, where the SHALSTAB model determines the stability-erosion of the slopes of the volcano, and the morphometric method to analyze the primary and secondary lahar currents, is a procedure that can help to prevent the destructive effects that they cause during these events as demonstrated in the same destroyed localities that were not mapped in the maps of existing hazards by CONRED. The field observations carried out by CONRED technicians and sent in the format of points in shp format, helped to calibrate and verify the results obtained from the presented methodology, as well as in understanding which of the three morphological parameters used had greater weight according to the scenario in the one that developed the lahars. This parameter corresponds to the TWI, the same one that determined that the lower areas of the streams are the ones that get wet and fill quickly overflowing and flooding nearby areas. The morphological model could obtain the primary and secondary lahar paths in all sectors near the volcano in a very short time, unlike the numerical models such as the LAHAR Z, which are applied in a single axis of a current or drainage. In this case, the paths of the lahars obtained are sufficient to Remote Sens. 2019, 11, 727 19 of 20 produce maps of the vulnerability reduction of civil structures and socio-economic systems including the preservation of human lives. This combination of procedures can be applied in any volcanic landscape based on digital models of high-resolution elevation, obtaining very good agreement with the paths of the real flows.

Author Contributions: The research work presented in this paper is a collaborative development by both authors. Conceptualization, M.C. and A.M.; Methodology, M.C. and A.M.; Software, M.C.; Validation, M.C. and A.M.; Formal Analysis, M.C.; Investigation, M.C. and A.M.; Resources, M.C.; Data Curation, M.C.; Writing-Original Draft Preparation, M.C. and A.M.; Writing-Review and Editing, A.M.; Visualization, Supervision: A.M.; Project Administration, A.M.; Funding Acquisition, A.M. Funding: This research received no external funding. Acknowledgments: This research was funded by projects Junta Castilla y León SA044G18 and projects from the Ministry of Economy and Competitiveness CGL2015-67169-P and CGL2015-69919-R. Conﬂicts of Interest: The authors declare no conﬂict of interest.

References

1. Almeida, S.; Daniel, S.; Daniel, A. Definition, Primary and Secondary Lahars, Flow Types, Behavior, afectation And Hazard Monitoring. Geophysical Institute National Polytechnic School, 2017. Available online: https://www.igepn.edu.ec/publicaciones-para-la-comunidad/comunidad-eng/19268-triptych- lahars-mud-flows/file (accessed on 21 February 2019). 2. National Coordinator for Disaster Reduction CONRED. Situation Report No. 1 (as of 04 June 2018). Internal Report. Available online: https://reliefweb.int/sites/reliefweb.int/files/resources/2018-06-04%20GT% 20SITREP%20-%20Volcanic%20Activity%20%28ENG%29.pdf (accessed on 21 February 2019). 3. Department of Vulcanology, INSIVUMEH. Preliminary Map of Hazard Scenarios by Pyroclastic Flows, after the Eruption of June 3, 2018. Available online: http://www.insivumeh.gob.gt/?page_id=70 (accessed on 21 February 2019). 4. National Coordinator for Disaster Reduction CONRED. Informative Bulletin No. 3162018—The Drop of Lahares Is Monitored on The Fuego Volcano. 2018. Available online: https: //conred.gob.gt/site/Boletin-Informativo-3162018andhttps://conred.gob.gt/www/index.php?option= com_content&view=article&id=7076&catid=37&Itemid=1010 (accessed on 21 February 2019). 5. León, X. Vulnerability Assessment Associated with the Hazard of the Fuego Volcano. Internal Study. Universidad de San Carlos de Guatemala, 2012. Available online: http://biblioteca.usac.edu.gt/tesis/02/ 02_3433.pdf (accessed on 24 February 2019). 6. Schilling, S.P. TI-LAHARZ: GIS Programs for Automated Mapping of Lahar-Inundation Hazard Zones. USGS Publications Warehouse, 1998. Available online: http://pubs.er.usgs.gov/publication/ofr98638 (accessed on 25 February 2019). [CrossRef] 7. Dietrich, W.E.; Bellugi, D.; Real de Asua, R. Validation of the Shallow Landslide Model, SHALSTAB, for forest management. Land Use and Watersheds: Human Influence on Hydrology and Geomorphology in Urban and Forest Areas. Water Sci. Appl. 2001, 2, 195–227. Available online: https://www.researchgate.net/publication/ 284902139_Validation_of_the_Shallow_Landslide_Model_SHALSTAB_for_forest_management (accessed on 25 February 2019). [CrossRef] 8. Montoya, S. Tutorial to Download Sentinel 2 Images (10 m Resolution) and Process Them in QGIS. GIDAHATARI, 2017. Available online: http://gidahatari.com/ih-es/tutorial-para-descargar-imgenes- sentinel-2-resolucin-10-m-y-procesarlas-en-qgis (accessed on 25 February 2019). 9. Zeni, G.; Pepe, A.; Zhao, Q.; Bonano, M.; Gao, W.; Li, X.; Ding, X. A differential SAR interferometry (DInSAR) investigation of the deformation affecting the coastal reclaimed areas of the Shangai megacity. In Proceedings of the International Geoscience and Remote Sensing Symposium (IGARSS), Quebec City, QC, Canada, 13–18 July 2014; pp. 482–485. [CrossRef] 10. The Copernicus Open Access Hub. The Open Access Hub Provides Complete, Free and Open Access to Sentinel-1, Sentinel-2, Sentinel-3 and Sentinel-5P User Products. Available online: https://scihub.copernicus. eu/userguide/ (accessed on 21 February 2019). Remote Sens. 2019, 11, 727 20 of 20

11. Foumelis, M.; Delgado Blasco, J.M.; Desnos, Y.-L.; Engdahl, M.; Fernández, D.; Veci, L.; Lu, J.; Wong, C. ESA SNAP—StaMPS Integrated processing for Sentinel-1 Persistent Scatterer Interferometry. 2018. Available online: https://www.researchgate.net/publication/327253039_ESA_SNAP_-_StaMPS_ Integrated_processing_for_Sentinel-1_Persistent_Scatterer_Interferometry (accessed on 23 February 2019). 12. Martins, T.; Vieira, B.; Fernandes, N.; Oka-Fiori, C.; Montgomery, D.R. Application of the SHALSTAB model for the identification of areas susceptible to landslides: Brazilian case studies. Revista de Geomorfologie 2017, 19, 136–144. 13. Panagos, P.; Borrelli, P.; Meusburger, K. A New European Slope Length and Steepness Factor (LS-Factor) for Modeling Soil Erosion by Water. Geosciences 2015, 5, 117–126. [CrossRef] 14. Zimmermann, M.; Mani, P.; Romang, H. Magnitude-frequency aspects of alpine debris flows. Eclogae Geol. Helv. 1997, 90, 415–420. Available online: https://www.e-periodica.ch/cntmng?pid=egh-001:1997:90::738 (accessed on 21 February 2019). 15. Carmona-Álvarez, J.E.; Ruge-Cárdenas, J.C. Análisis de las correlaciones existentes del ángulo de fricción efectivo para suelos del piedemonte oriental de Bogotá usando ensayos in situ. Tecno Lógicas 2015, 18, 93–104. Available online: http://www.scielo.org.co/pdf/teclo/v18n35/v18n35a09.pdf (accessed on 23 February 2019). [CrossRef] 16. Dietrich, W.E.; Montgomery, D.R. A Digital Terrain Model for Mapping Shallow Landslide Potential. University of California-University of Washington, 1998. Published as a Technical Report by NCASI. Available online: http://calm.geo.berkeley.edu/geomorph/shalstab/index.htm (accessed on 23 February 2019). 17. O’Loughlin, E.M. Prediction of surface saturation zones in natural catchments by topographic analysis. Water Resour. Res. 1986, 22, 794–804. Available online: https://www.scirp.org/(S(i43dyn45teexjx455qlt3d2q) )/reference/ReferencesPapers.aspx?ReferenceID=729751 (accessed on 21 February 2019). [CrossRef] 18. Sebastiano, T.; Cavalli, M. Topography-based flow-directional roughness: potential and challenges. Earth Surf. Dyn. 2016, 4, 343–358. Available online: https://www.earth-surf-dynam.net/4/343/2016/ esurf-4-343-2016.pdf (accessed on 26 February 2019). 19. Beven, K.J.; Kirkby, M.J. A physically based variable contributing area model of basin hydrology. Hydrol. Sci. Bull. 1979, 24, 43–69. Available online: https://www.tandfonline.com/doi/abs/10.1080/02626667909491834 (accessed on 26 February 2019). [CrossRef] 20. Cando, M.; Martínez-Graña, A. Numerical modeling of flow patterns applied to the analysis of the susceptibility to movements of the ground. Geosciences 2018, 8, 340. Available online: https://www. mdpi.com/2076-3263/8/9/340 (accessed on 26 February 2019). [CrossRef] 21. Hojati, M.; Mokarram, M. Determination of a Topographic Wetness Index using high resolution Digital Elevation Models. Eur. J. Geogr. 2016, 7, 41–52. Available online: http://www.eurogeographyjournal.eu/ articles/4.%20DETERMINATION%20OF%20A%20TOPOGRAPHIC%20WETNESS%20INDEX%20USING% 20HIGH%20RESOLUTION%20DIGITAL%20ELEVATION%20MODELS.pdf (accessed on 25 February 2019). 22. Ebmeier, S.; Biggs, J.; Mather, T.; Amelung, F. Applicability of InSAR to tropical volcanoes: Insights from Central America. Geol. Soc. Lond. Spec. Publ. 2013, 380, 15–37. Available online: https://www.researchgate.net/publication/259130455_Applicability_of_InSAR_to_ tropical_volcanoes_Insights_from_Central_America (accessed on 21 February 2019). [CrossRef] 23. Baumann, V.; Bonadonna, C.; Cuomo, S.; Moscariello, M.; Manzella, I. Slope stability models for rainfall-induced lahars during long-lasting eruptions. J. Volcanol. Geotherm. Res. 2018. Available online: https://www.researchgate.net/publication/326139979_Slope_stability_models_for_ rainfall-induced_lahars_during_long-lasting_eruptions (accessed on 25 February 2019). [CrossRef] 24. Vidal Montes, R.; Martínez-Graña, A.M.; Martínez Catalán, J.R.; Ayarza, P.; Sánchez San Román, F.J. Vulnerability to groundwater contamination, SW Salamanca, Spain. J. Maps 2016, 12, 147–155. [CrossRef] 25. Martínez-Graña, A.M.; Goy, J.L.; Zazo, C. Geomorphological applications for susceptibility mapping of landslides in natural parks. Environ. Eng. Manag. J. 2016, 15, 1–12. [CrossRef] 26. Santos-Francés, F.; Martínez-Graña, A.; Rojo, P.A.; Sánchez, A.G. Geochemical background and baseline values determination and spatial distribution of heavy metal pollution in soils of the Andes mountain range (Cajamarca-Huancavelica, Peru). Int. J. Environ. Res. Public Health 2017, 14, 859. [CrossRef][PubMed]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).